That's a nice idea. With some more potential of tightening up things, enough space should be availbale...
I was hesitant to go too thin on the track width and tried to use .4 mm most of the time. Only when I had to route between IC pins I reduced that segement's width a little bit to leave enough space.
Not making the tracks too thin was such a common advice that I read so I got really careful there - on the otherhand, with the signal levels we're dealing with here, I also think .25 mm should be a problem, in theory.
I'll finish up this initial build, see if there are any blunders and then I'll think about where this goes 🙂
Trust me 0.25mm trace is not thin, for most analogue layout I do (and that is lot) 0.20mm (8 thou) track and gap is standard with a 0.60mm Via 0.3mm finished via hole. 0.25mm is perfect for this type of design, do not bother changing the widths, it makes for a messy layout and is not required. For critical signals such as reference voltages I would use 0.50mm, and for PSU stubs down to power planes I would use a track width the same as the via diameter. As said for the extra cost a 4 layer board has many advantages, a contiguous ground plane on layer two, and better power distribution. Do not try and route star grounds on a PCB, it does not work very well at all, thin 0V traces add inductance and resistance. AS to the actual blocks, lay out feedback loops first as small as possible, then work out from there. All op-amp blocks are pretty standard... Once you've done a few it becomes second nature.
Thanks for the practical advice thisis good to know for either a next revision or the next project. I guess this is the kind of stuff that you really need expirience to pick up on beceause just from reading about it you get so many conflicting opinions. It's a fun journey though and I am happy that I got this far.
One channel is almost working but something is still odd about the tone control - it produces strange clicking when you try to dial in the boost or cut. Not a continous distortion but more like clicks/dropouts at a very low frequency (like one or two times per second). The other channel is still silent probably some bad solder point. So I have a bit of debugging to do now...
Strangely enough, the channel that is almost working has ~5.5VDC at the output which makes no sense at all since there are the 100uF C45/C46 decoupling caps right before the output. I wonder what I messed up there ;-)
One channel is almost working but something is still odd about the tone control - it produces strange clicking when you try to dial in the boost or cut. Not a continous distortion but more like clicks/dropouts at a very low frequency (like one or two times per second). The other channel is still silent probably some bad solder point. So I have a bit of debugging to do now...
Strangely enough, the channel that is almost working has ~5.5VDC at the output which makes no sense at all since there are the 100uF C45/C46 decoupling caps right before the output. I wonder what I messed up there ;-)
Thanks for asking...
I had to take a bit of a pause because I was very busy at my job. I found a major bug in the PCB because I had placed some of the OpAmps "upside-down" when putting the schematics in KiCad so inverted and non-inverted input are swapped on the PCB. This happened because the default op-amp symbol in KiCad has the non-inverted input at the top (which is not the convention I'm used to) and I did not pay enough attention.
My solution now is that I have designed a little mini adapter PCB that should fit on the SOIC-8 pads and swaps the inputs and I'm trying to solder that onto the PCB and put the OpAmps onto that. I just ordered the adapters and crossing my fingers that this will work.
Other measurements I was able to do looked promising so I hope there won't be other bugs of similar criticality...
Ouch, having to do an interposters is a pain... Though due to some components being on silly lead times we are doing more than usual, I have one on my desk... One tip, if you can use a higher temp solder to fix the interposter to the main PCB, then a lower temp solder to place the chip, in can make life a bit easier.
Something else to be aware of is orientation, not as bad with an 8 pin device, but worth double checking especially if you use a schematic, the bottom component is mirrored...
Something else to be aware of is orientation, not as bad with an 8 pin device, but worth double checking especially if you use a schematic, the bottom component is mirrored...
Thanks for the tips. I didn't know "interposer" was the official term for this sort of thing.
I think I'm good on orientation. I specifically created a mirrored footprint for the underside and created the PCB without going through a schematic for fear of confusing things again. Just approaching this one visually seemed to work better for my brain.
There's also another funny mistake I made where I picked a generic symbol for the BJTs in the power supply that had base and collector swapped. Q_PNP_BCE is the first symbol that shows up when you type PNP or NPN in KiCad and I just went with that not paying attention to the fact that it has the base on pin 1 and the collector on pin 2. Obviously, I could just bend the device's leads in place. Still, another lessons learned to pay attention to detail.
The poor thing now has to reside in a permanent twisted roots yoga pose 🙂
I think I'm good on orientation. I specifically created a mirrored footprint for the underside and created the PCB without going through a schematic for fear of confusing things again. Just approaching this one visually seemed to work better for my brain.
There's also another funny mistake I made where I picked a generic symbol for the BJTs in the power supply that had base and collector swapped. Q_PNP_BCE is the first symbol that shows up when you type PNP or NPN in KiCad and I just went with that not paying attention to the fact that it has the base on pin 1 and the collector on pin 2. Obviously, I could just bend the device's leads in place. Still, another lessons learned to pay attention to detail.
The poor thing now has to reside in a permanent twisted roots yoga pose 🙂
Last edited:
Not yet, work is super-busy at the moment so I don't get much time at the workbench. I've fitted the first of the interposers and I wish I had designed them just a wee-bit thinner to easier heat up the pads on the mainboard (I don't have a hot-air station) but I think I can make it work...